LOADING加载中
46 个模型
| 排名 | 模型 | 评分 | 评分下界 | 评分上界 | 方差 | 投票数 | 分类 | 源数据更新 |
|---|---|---|---|---|---|---|---|---|
| #1 | Claude Fable 5.1 (Max)anthropic · Proprietary | 1,499.7 | 1,529.99 | 59.73 | 1,453 | overall | 2026-11-09 | |
| #2 | Claude Opus 5 (High)anthropic · Proprietary | 1,503.03 | 1,514.3 | 8.27 | 16,189 | overall | 2026-11-09 | |
| #3 | Claude Opus 4.6 (High)anthropic · Proprietary | 1,501.25 | 1,513.96 | 10.52 | 29,456 | overall | 2026-11-09 | |
| #4 | Claude Opus 4.6anthropic · Proprietary | 1,501 | 1,511.44 | 7.1 | 47,264 | overall | 2026-11-09 | |
| #5 | Claude Fable 5 (High)anthropic · Proprietary | 1,489.96 | 1,503.25 | 11.5 | 12,357 | overall | 2026-11-09 | |
| #6 | Claude Opus 4.7 (High)anthropic · Proprietary | 1,489.26 | 1,502.11 | 10.73 | 23,327 | overall | 2026-11-09 | |
| #7 | Claude Opus 4.7anthropic · Proprietary | 1,488.4 | 1,501.17 | 10.62 | 23,628 | overall | 2026-11-09 | |
| #8 | GPT-5.5 (High)openai · Proprietary | 1,478.94 | 1,489.64 | 7.45 | 27,324 | overall | 2026-11-09 | |
| #9 | Gemini 3.8 Flash (High)google · Proprietary | 1,473.98 | 1,489.86 | 16.41 | 6,950 | overall | 2026-11-09 | |
| #10 | Claude Sonnet 4.6anthropic · Proprietary | 1,476.03 | 1,487.72 | 8.91 | 61,055 | overall | 2026-11-09 | |
| #11 | GPT-5.5openai · Proprietary | 1,475.59 | 1,486.2 | 7.33 | 27,751 | overall | 2026-11-09 | |
| #12 | GPT-5.6 Sol (xHigh)openai · Proprietary | 1,473.24 | 1,486.84 | 12.03 | 9,649 | overall | 2026-11-09 | |
| #13 | Gemini 3.7 Flash (High)google · Proprietary | 1,468.87 | 1,484.07 | 15.04 | 7,680 | overall | 2026-11-09 | |
| #14 | Claude Opus 4.8 (High)anthropic · Proprietary | 1,467.7 | 1,481.4 | 12.22 | 12,069 | overall | 2026-11-09 | |
| #15 | GPT-5.4openai · Proprietary | 1,466.99 | 1,478.91 | 9.25 | 35,183 | overall | 2026-11-09 | |
| #16 | Muse Spark 1.3 (Max)meta · Proprietary | 1,465.14 | 1,480.7 | 15.74 | 7,205 | overall | 2026-11-09 | |
| #17 | GPT-5.6 Terra (xHigh)openai · Proprietary | 1,461.23 | 1,474.09 | 10.76 | 11,412 | overall | 2026-11-09 | |
| #18 | GPT-6 Astra (Max)openai · Proprietary | 1,458.67 | 1,475.84 | 19.19 | 5,723 | overall | 2026-11-09 | |
| #19 | Claude Opus 4.5 (20251101)anthropic · Proprietary | 1,455.5 | 1,475.92 | 27.14 | 8,304 | overall | 2026-11-09 | |
| #20 | Claude Opus 4.8anthropic · Proprietary | 1,458.7 | 1,471.4 | 10.5 | 14,729 | overall | 2026-11-09 | |
| #21 | Muse Spark 1.1meta · Proprietary | 1,456.37 | 1,470.56 | 13.1 | 8,934 | overall | 2026-11-09 | |
| #22 | Gemini 3.5 Flash (High)google · Proprietary | 1,452.61 | 1,470.92 | 21.82 | 4,248 | overall | 2026-11-09 | |
| #23 | Gemini 3.5 Flash (Medium)google · Proprietary | 1,452.86 | 1,468.42 | 15.75 | 6,835 | overall | 2026-11-09 | |
| #24 | Claude Sonnet 5 (High)anthropic · Proprietary | 1,452.97 | 1,466.11 | 11.23 | 11,044 | overall | 2026-11-09 | |
| #25 | Gemini 3.6 Flash (High)google · Proprietary | 1,445.43 | 1,466.82 | 29.79 | 3,031 | overall | 2026-11-09 | |
| #26 | GPT-5.6 Luna (xHigh)openai · Proprietary | 1,446.78 | 1,460.82 | 12.84 | 8,832 | overall | 2026-11-09 | |
| #27 | Grok 4.5xai · Proprietary | 1,443.92 | 1,459.07 | 14.94 | 7,447 | overall | 2026-11-09 | |
| #28 | Kimi K2.6moonshot · Modified MIT | 1,442.91 | 1,457.99 | 14.81 | 11,809 | overall | 2026-11-09 | |
| #29 | Claude Sonnet 4.5 (20250929)anthropic · Proprietary | 1,443.26 | 1,455.42 | 9.61 | 32,916 | overall | 2026-11-09 | |
| #30 | Muse Sparkmeta · Proprietary | 1,425.44 | 1,461.67 | 85.41 | 1,116 | overall | 2026-11-09 | |
| #31 | Gemini 3.1 Pro Previewgoogle · Proprietary | 1,438.06 | 1,448.45 | 7.02 | 52,244 | overall | 2026-11-09 | |
| #32 | Grok 4.6 (High)xai · Proprietary | 1,430.34 | 1,445.83 | 15.6 | 7,669 | overall | 2026-11-09 | |
| #33 | Qwen3.7-Plusalibaba · Proprietary | 1,429.97 | 1,444.91 | 14.52 | 7,426 | overall | 2026-11-09 | |
| #34 | MiniMax-M3minimax · MiniMax Community License | 1,428.81 | 1,445.02 | 17.1 | 6,617 | overall | 2026-11-09 | |
| #35 | Gemma 4 31Bgoogle · Apache 2.0 | 1,427.61 | 1,441.06 | 11.77 | 15,431 | overall | 2026-11-09 | |
| #36 | Gemini 3 Progoogle · Proprietary | 1,425.14 | 1,442.59 | 19.83 | 11,338 | overall | 2026-11-09 | |
| #37 | Kimi K2.5 (Thinking)moonshot · Modified MIT | 1,424 | 1,437.35 | 11.6 | 22,601 | overall | 2026-11-09 | |
| #38 | Gemini 2.5 Progoogle · Proprietary | 1,416.66 | 1,428.92 | 9.78 | 26,635 | overall | 2026-11-09 | |
| #39 | Claude Haiku 4.5 (20251001)anthropic · Proprietary | 1,415.09 | 1,425.72 | 7.35 | 39,423 | overall | 2026-11-09 | |
| #40 | GLM-5V-Turbozai · Proprietary | 1,411.37 | 1,426.16 | 14.23 | 9,136 | overall | 2026-11-09 | |
| #41 | Grok 4.20 Beta (Reasoning, 0309)xai · Proprietary | 1,410.21 | 1,423.98 | 12.35 | 21,926 | overall | 2026-11-09 | |
| #42 | Gemini 3 Flashgoogle · Proprietary | 1,404.96 | 1,423.62 | 22.65 | 7,538 | overall | 2026-11-09 | |
| #43 | GPT-5.2 (High)openai · Proprietary | 1,394.65 | 1,413.11 | 22.17 | 7,552 | overall | 2026-11-09 | |
| #44 | GPT-5.5 Instantopenai · Proprietary | 1,395.42 | 1,411.06 | 15.93 | 8,834 | overall | 2026-11-09 | |
| #45 | GPT-5.1openai · Proprietary | 1,393.9 | 1,412.13 | 21.64 | 8,736 | overall | 2026-11-09 | |
| #46 | GPT-5.2openai · Proprietary | 1,396.27 | 1,407.9 | 8.81 | 29,726 | overall | 2026-11-09 |
用113个从零编写的长程真实软件工程任务测评编程Agent，同时刻画准确率(Pass@1)与真实花费/步数的性价比曲线，避免题目训练污染。
在新标签页打开独立跑分，不依赖厂商自报数据；综合 Intelligence Index 覆盖推理/编程/科学/知识，并联动速度、价格、上下文窗口等工程指标。
在新标签页打开每月更新全新题目防污染，全自动客观判分（唯一正确答案），聚焦数学/代码/推理/数据分析，学术圈公认权威。
在新标签页打开中文原创题库，深度贴合中文语境与政策文化，覆盖数学推理、代码、幻觉控制、智能体规划六大能力。
在新标签页打开上海AI实验室出品，20余家企业共建，五大能力维度、70+ 数据集，中英文模型均覆盖，并含真人投票的 Compass Arena。
在新标签页打开本页对战数据的官方原始来源，累计投票超过700万次，覆盖370+模型，逐票更新，界面本身也值得参考。
在新标签页打开